[Air-l] counting google hits

[Thomas Koenig]

Heidi,
Scrive Heidi haLevi <heidi at processing.co.il>:

> > Just like the results SPSS delivers depends, of course, on the data
> file,
> > which you can manipulate as you like it. Does that make SPSS an
> unreliable
> > tool?
>
> uhm...
> may i nonchalantly point out here the vast difference between changing
> the
> content of the data file and changing the content of the query run on the
> data file?

That's what I meant, I should have been more clear about it: You can have 8
potential indicators for class location of a person, and depending on the
statistical method you are using, and the selction and/or combination of
the indicators, your results will vary. If you choose bad indicators (such
as "sex (m/f)" for class location in SPSS or "MacDonalds" for McDonald's'
popularity in googlefight, your results might become misleading.

> heh - kinda neat to consider the internet under that title: an open data
> file.

Err. How else would you conceptualize it?

> thus - as with every statistical analysis - the form of the query should
> be
> carefully matched to the research question as well as to the form of the
> data and of the file. in this case, the data sample is actually the
> population/field, and google's particular form of the 'data file' (or the
> determination of which variables are accessible for measurement and how)
> is
> closed and not within our reach, and also determines the possible
> "statistic
> analyses" - which are very primitive (though certainly always
> statistically
> significant).
>
> an interesting situation, to be approached and used with caution.

I hate to write that: I agree.

Well, except, why should all statistical analyses on the Google "corpus"
always be primitive?

BTW: Besides google fight there is also googleshare
(http://www.stevenberlinjohnson.com/movabletype/archives/000011.html):
Unfortunately, there only seems to be a Flash implementation for Internet
Explorer right now: http://www.rundogrun.com/Samples/MindShare.HTML

Thomas

--
thomas koenig, ph.d.
department of social sciences, loughborough university, u.k.
http://www.lboro.ac.uk/research/mmethods/staff/thomas/index.html